Labelling this “game speed” is very unintuitive to me. I would expect that enabling “flexible” means you don’t need to pick one of the options, since you are flexible on time.
It should be called “timing” or something and when enabled the fisher/byo-yomi buttons be merged and just look like the Lichess ones but with the approx game duration and a name (blitz, rapid, normal or whatever)
Or if that’s too radical then at least make selecting one also highlight the other in the pair.
Noting the discussion on this I’d suggest moving this option to settings. If it’s a thing that is “hardwired” for people that individuals either never want to play handicap or don’t care then that would be a good way to retain the option but clean the play page. And there’s always custom if people want a one off different game to that in their settings.
(And the flexible button could also be a setting with a better label about “flexible on automatch clock method” or something)
I agree with Uber here. The purpose of those timings is to know if I have time to play this game at that setting. So you only need to show the approx upper limit and it can be rounded.
So for 9x9 it’s ~5, <10 and <15
For 13 it’s <15, <20/25 and <30/35
And for 19 it’s <15, <30/35 and <50
On mobile it’s annoying because it’s way too big. It should just be a thumbnail to the right of the size options.
And anyway, it shouldn’t be a boring empty board but a live random game from the watch page …
Yep “aiming fairness” much better as “difficulty balance”.
That’s like two opposite ways to ask, attracting by the fairness or by the difficulty.
Not my opinion, when I go to play I am preparing myself with an empty board. So showing games running may be confusing (especially for players not familiar with UI) or stressing like @genbeart said.
Fwiw I like the empty board, it makes it obvious which size is selected. If it’s just the thin blue outline of the button, I’ll probably start some games on the wrong size by mistake. I’d even suggest to also put a symbol in the middle of it to indicate the selected time settings (a bullet for bullet/ lightning bolt for blitz/ clock for normal/ turtle for correspondence). Maybe one could also put the maximum handicap stones for the selected rank range on the board, but that might be confusing.
This is a culture difference. Both sides exists, some learned the game through climbing the handicap stones, some not at all. Keeping both options seems a bit unavoidable.
Sounds an important point to me too.
Something like 2 separate switches?
[ fairness ] ____ [ difficulty ]
Last but not least, I hope you can make multiple selection at the same time (didn’t check the UI yet) so that I can wait for different types of game. Talking about all the settings btw
And for automatch spirit we may have a big simple button written “just have a live game” somewhere on top which would preselect all but not the correspondance one.
I agree with Uberdude about reducing the number of options for timing. If people really want a specific timing option they can make a custom game.
If all the buttons are staying perhaps they could be individually toggle-able instead of having the “Flexible” switch? Then I can tick a bunch of them that fits my current time slot. (Same for board size but UX becomes tricky then)
In support of this, the results of the poll seemed to lean in favor of a single option - admittedly not unanimous which to choose, but “split pool” only had 15%.
I think I like when it says 10-15 mins etc for the times. It feels like it’s slightly better than just telling me the longest possible game that almost never happens.
An average and a reasonable upper limit should be good enough for people to get an idea of how long they have to commit.
If I play fast no matter what, maybe we hit that average time, if both me and my opponent use a lot of our time and there’s a number of kos and so on, then maybe we push into the upper limit part.
I feel like it’s actually not that many - it’s already kind of reduced to the core most popular settings.
Blitz , rapid , live , correspondence.
Byo-yomi was very popular but it was also the default except I think for blitz, when there was a decent fraction of custom Fischer games especially for 9x9.
I don’t think there’s a clean way to remove either byo-yomi or Fischer without annoying a decent number of people, given some of the polls.
I think if one side cleanly won the poll you’d know what to do, but because no side really did, and Fischer definitely didn’t beat byo-yomi in the other poll 60-40 say, I don’t think you can conclude much.
Really it’s more like that some of the byo-yomi and Fischer people picked another option say when given more options (split unanimous etc)
The results of that poll were dirty. I also voted for the top option even though I prefer Fischer because I wanted just one option. So did many others. That’s why I made my poll in response where we can see 85% of people want just one timing option, with the number of people preferring Fischer being 50% higher than those that prefer byo-yomi. Much less controversial than many things we’ve added.
Unfortunately I think you need an even cleaner poll, because you didn’t really address Fischer vs Byo-yomi any better because again it’s muddied by another two options to say to have a uniform time option either with or without preference, or to have a split.
So you can’t really tell who actually wanted Fischer or Byo-yomi out of the three non specific options taking like 50% of the vote.
I’m thinking the thing to do is to launch with the two choices and after a bit of time we’ll look at the stats to see if there’s a significant skew to one or the other. Flexible matching is on by default so there shouldn’t be a notable effect by the split pools (since you’re in both). I worry that that poll is going to be a bit skewed since it has the fear factor about the pool size, but I think that’s not as big of a deal with flexible on by default (which is why it is.)
Edit: For what it’s worth I also really would like just one option, but we’re at this cross roads where Byo-Yomi is the de facto standard but there is a significant push by a lot of players and Go organizations to switch to Fischer, so I think we need to give it a good chance, but I also don’t know that there’s a critical mass enough to just switch to it from Byo-Yomi.
It’s also the case that automatch users just want a game quicker.
So if there’s any improvements to speed up the auto matching algorithm, I bet that would also make a difference to whether people care if there’s a X options or Y pools.
We have reason to think that maybe there could be some improvements possible? The latest WSC challenge we had some comments about the automatch maybe being a bit slow at times
I too when testing time settings on beta, to record which ones were being used by automatch, found the automatch to be a bit slow, given I was using several accounts that should’ve matched with each other and I was 100% the only accounts in the queue
So if there’s any intentional waiting times, like “there’s is a match in the pool but we’re waiting X seconds or minutes for a better match to come along first”, as part of the algorithm, then that could be one point of optimisation for example.
Especially when there are flexible options like a rank range or handicap or time settings:
if you want a fairly specific matchup change your settings,
If you have it fairly flexible like up to 9 stones, a huge rank range any time control, it should probably match quickly over finding someone within one rank and waiting for it?
Edit: I’m not saying it’s bad, but rather that even a little optimisation if possible probably goes a long way?
There is a pooling delay of sorts. We don’t take the first match but rather look for a great match (and will accept it pretty quickly if it’s available) - this would be an opponent that’s your same rank. The larger the rank difference the more time we’ll wait before pairing those two.
If two players have played each other recently we exclude them from matching again for some period of time. Typing that out now though I realize that should just be an increase on the match delay not a hard block, so I’m going to change that. I checked the code and it’s a suppression on match time but not a hard block, so while I might tweak the suppression level, functionally it’s doing what we want I think.
I’d say I ran into this one on beta, because sometimes a pair of accounts would match quickly, but when I cancelled the game (I just wanted to read off the settings) they wouldn’t match up again for quite a while if even.
This sounds like it makes sense when the pool is very active, but at quieter times of the day, maybe this harms players that are queuing up with flexible settings?
I guess what I’m thinking is that if you have uniform activity across all rankings then waiting for a closer match makes sense, but maybe if it’s a multimodal distribution at certain times of the day maybe certain ranks with flexible settings would suffer. I’ll try draw a bad picture
So uniform everything is good (or even like a uniform base where the minimum players in a time window are like 3,4,… so there’s always a quick enough match)
But high variance, multimodal where the minimums average like <2 players over some window of time, could be a problem and even changing settings to the most flexible won’t speed up the matching time?
Like it would suck to be a 15kyu or a 5kyu in the second picture with strict settings (at this time of day say), but changing to flexible wouldn’t necessarily help if there’s a delay corresponding to rank difference.
Indeed. I am adding a number of stats but I’m not sure I want to tackle tuning delays based on those dynamic stats just yet in this patch. I’m thinking in the interm though maybe the thing to do is to reduce the delays based on rank since that’s an approximatable curve (so the radius of acceptable games expands quicker for a 5d than for a 12k).
No, ability to choose weaker or stronger opponents in even games is needed. Win streaks are boring and loss streaks are painful. So when I have more wins than loses, I play equal and stronger opponents only. When I have equal number of wins and loses, I hide ranks of opponents and play game without rank limits. Its fun to have no idea if your current opponent is ddk or dan.
I (5k) was ahead vs 3d on move 62 before getting crushed. Against 6d I was ahead on move 37. (by Kata graphs)
So I can train my fuseki without caring about game result. Handicap stones would interfere with fuseki training.
And when I have more loses than wins I can relax by playing weaker and equal only.
If I always had 50% chance to win in every game I would not be able to relax.
) but I think it’s pretty clear that automatch users want a less fragmented player pool.
